April 28, 2015 12:39 WSPC - Proceedings Trim Size: 9in x 6in clawar15brehman 1 Design of a Hydraulically Actuated Arm for a Quadruped Robot Bilal Ur Rehman * , Michele Focchi, Marco Frigerio, Jake Goldsmith, Darwin G. Caldwell and Claudio Semini Department of Advanced Robotics, Istituto Italiano di Tecnologia, via Morego, 30, 16163 Genova, Italy * E-mail: [email protected]www.iit.it A common disadvantage of quadruped robots is that they are often limited to load carrying or observation tasks, due to their lack of manipulation capabil- ity. To remove this limitation, arms can be added to the body of the robot, enabling manipulation and providing assistance to the robot during body sta- bilization. However, a suitable arm for a quadruped platform requires specific features which might not all be available in off-the-shelf manipulators (e.g. speed, torque-controlled, light-weight, compact, without external control unit). In this paper, we present a systematic approach to design a robotic arm tailored for an 80kg quadruped robot. A full robot with arms and legs (aiming for a cen- taur -style robot) was simulated performing a range of “representative” tasks to estimate joint torques and velocities. This data was then extensively used to select the design parameters, such as the joint actuators to develop a novel, compact (0.743m fully extended), light-weight (12.5kg), and fast (maximum 4m/s no-load speed at end-effector) hydraulically actuated robotic arm with 6 torque-controlled degrees of freedom. The enclosed video presents preliminary experimental results. Keywords : Manipulator Design, Hydraulically Actuation, Torque Controlled, Hydraulic Quadruped, Multi-legged Robot 1. Introduction How does a robot transverse highly uneven terrain? And what does it do when it gets to its destination? This is an area that is expected to be cov- ered by legged robots. On the whole, quadrupeds have the advantage (over bipeds) of improved stability, whilst not becoming overly complex (like hexapods). Traditionally quadrupeds have been limited to load carrying or observation tasks, as they have no manipulation ability. This paper presents a “best-of-both-worlds” approach, by a bespoke arm system which can be mounted on the hydraulic quadruped robot HyQ, 1 in a single or bimanual
8
Embed
Design of a Hydraulically Actuated Arm for a Quadruped Robot · the base, which reduces the arm inertia) with a constant torque output. The elbow joint is actuated by a hydraulic
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
April 28, 2015 12:39 WSPC - Proceedings Trim Size: 9in x 6in clawar15brehman
1
Design of a Hydraulically Actuated Arm for a Quadruped Robot
Bilal Ur Rehman∗, Michele Focchi, Marco Frigerio, Jake Goldsmith, Darwin G.
Caldwell and Claudio Semini
Department of Advanced Robotics, Istituto Italiano di Tecnologia,
A common disadvantage of quadruped robots is that they are often limited toload carrying or observation tasks, due to their lack of manipulation capabil-ity. To remove this limitation, arms can be added to the body of the robot,
enabling manipulation and providing assistance to the robot during body sta-bilization. However, a suitable arm for a quadruped platform requires specific
features which might not all be available in off-the-shelf manipulators (e.g.speed, torque-controlled, light-weight, compact, without external control unit).In this paper, we present a systematic approach to design a robotic arm tailoredfor an 80kg quadruped robot. A full robot with arms and legs (aiming for a cen-
taur -style robot) was simulated performing a range of “representative” tasksto estimate joint torques and velocities. This data was then extensively usedto select the design parameters, such as the joint actuators to develop a novel,
compact (0.743m fully extended), light-weight (12.5kg), and fast (maximum4m/s no-load speed at end-effector) hydraulically actuated robotic arm with 6
torque-controlled degrees of freedom. The enclosed video presents preliminaryexperimental results.
Wrist Rotation (WR), Wrist Flexion/Extension (WFE) All the joints are shown atzero configuration
3. Simulation
This section presents the simulation results for a set of “representative”
tasks allowing the estimation of torque and velocity profiles for actuator
selection (see subsection. 3.1). The robot model used in the simulations is
a floating base quadruped robot with two arms (see Fig. 4 of Centaur). We
used SL10 that is a real-time simulation environment for rigid bodies, where
we implemented the controller as well. To build the dynamics for the simu-
lated robot, we used RobCoGen,11 a model based code generator to provide
kinematics and dynamics of articulated robots.12 Given the kinematic tree
of the robot and its inertia properties, RobCoGen automatically generates
forward/inverse dynamics and kinematic transforms targeting different soft-
ware platforms. The parameter for the HyQ robot are taken from the CAD
model. To calculate inertia properties for the robotic arm simulation (con-
sidering the design specifications given in section 2), we selected each link as
represented by an aluminum cylindrical link of mass 2kg, length 0.175[m],
diameter 0.075[m] with material density of 2700[kg/m3].
3.1. Representative tasks
We designed the “representative” task trajectories to be demanding in
terms of torque or velocity, for a single or a combination of joints. We
developed minimum jerk trajectories for the end-effector of the arm in the
Cartesian space (unless otherwise specified). These then resulted in the
motion of the arm joints. The remaining joints are kept in a default config-
uration. An impedance control law defined both for position and orientation
April 28, 2015 12:39 WSPC - Proceedings Trim Size: 9in x 6in clawar15brehman
5
Figure 4. Picture of the Centaur simulation during the pull-up (left) and push-up (right)tasks.
sets the virtual forces/torques (F ,T ) at the end-effector. We also simulated
the robot falling to estimate fall time. Fig. 5 summarizes required torque
and velocity plots for each joint for all simulated tasks. The explanation of
each simulated task is presented as follows:
(a) Lifting an object: This task simulates the centaur robot lifting
an object located at the end-effector when the arm is fully extended. It
demands high torque for shoulder and elbow joints. We set three different
kind of trajectories for shoulder joints: horizontal (SAA), vertical (SFE),
humerus rotation. Each scenario has been simulated with and without pay-
load which are two different estimations for maximum joint torque and
velocity, respectively. We set a conservative payload of 5kg at end-effector
moving at speed of about 1.5m/s for the shoulder joints. These trajectories
were also simulated without payload at a speed of 4m/s (three times faster
than the robot falling time).
SAA SFE HR EFE WR WFE
Max torq
ue [N
m]
0
20
40
60
80
100Required Torque
LOH under loadLOV under loadLHR under loadPull-upPush-up
SAA SFE HR EFE WR WFE
Max v
elo
city [ra
d/s
]
0
0.5
1
1.5
2
2.5
3Required Velocity
LOH without loadLOV without loadLHR without loadPull-upPush-up
(a) (b)
Figure 5. (a)Required torque and (b) velocity for lifting an object, biceps curl, pull-up and wall push-up tasks for each arm joint. LO:Lifting an Object, V:Vertical,H:Horizontal, HR:Humerus Rotation
(b) Pull up: This task demands a high torque output for the shoulder
and elbow joints. In this task an arm is holding a vertical beam and pulling
the robot torso (up to 0.3m) towards the beam while standing on a slope
of 0.5 rad inclination (Fig. 4 (left)). This task resembles opening a door or
April 28, 2015 12:39 WSPC - Proceedings Trim Size: 9in x 6in clawar15brehman
6
pulling an object.
(c) Wall push-up: The wall push-up task (Fig. 4(right)) demands high
torque for shoulder and elbow joints. For this task the robot is standing on
a 0.5 rad inclined slope and performing a push-up motion against a wall
which resembles to provide assistance to robot while climbing stairs or to
balance.
4. HyArm Design
The simulated tasks of Section 3, provided the required peak torques, and
velocities for each DoF. According to simulated data, we intended to select
commercially available hydraulic actuators which are as light weight and
compact as possible.
The HyArm joints are actuated with a combination of rotary and lin-
ear hydraulic actuators Fig. 2. The benefit of using this combination is
to achieve large joint ranges while still ensuring both a compact and light-
weight design. The HyArm shoulder joints (3DoF ) are equipped with rotary
motors to improve compactness (also keeping the shoulder CoM closer to
the base, which reduces the arm inertia) with a constant torque output.
The elbow joint is actuated by a hydraulic cylinder. A four bar linkage
(inspired by the excavator bucket joint) has been designed for this joint to
achieve a good trade-off between joint range-of-motion and output torque.
This choice has the advantage that the whole elbow assembly is part of
the upper arm. The wrist joints play an important rule in determining
end-effector position and orientation. For the WR joint we selected a ro-
tary actuator to achieve wider range-of-motion, while the WFE joint is
actuated by a cylinder. A standard lever mechanism provides the required
range-of-motion and torque for the WFE joint.
Referring to Fig. 2(right) (elbow and wrist assembly) using known ge-
ometric calculations (law of cosines) we developed the kinematic relation-
ship between the angles θi and the effective lever arm Leffi . The effective
lever arm is the quantity which allows to map joint angular velocity θiand torque τi into cylinder linear velocity xcyli = Leffi(θi)θi and force
Fcyli = L−1
effi(θi)τi, where i = 4, 6 represents the joint numbers (EFE
and WFE respectively). The HyArm is equipped with position encoder,
torque and force sensors (see Fig. 2(right)) to achieve torque control. Ta-
ble. 1 presents an overview of the arm specification. The enclosed video
(see section 6) presents preliminary experimental results. The HyArm is
demonstrating torque controlled capability to change joint impedance and
user and robot interaction while performing a continuous motion with dif-
ferent speeds.
April 28, 2015 12:39 WSPC - Proceedings Trim Size: 9in x 6in clawar15brehman
8
6. Appendix
The youtube link of simulation and real robot experiments (under torque
controlled): http://youtu.be/JhbHPZc-NGU
Bibliography
1. C. Semini, N. G. Tsagarakis, E. Guglielmino, M. Focchi, F. Cannella andD. G. Caldwell, Design of HyQ - a hydraulically and electrically actuatedquadruped robot, IMechE Part I: Journal of Systems and Control Engineer-ing 225, 2011.
2. V. Barasuol, J. Buchli, V. J. D. Negri, E. R. D. Pieri, D. G. Caldwell andC. Semini, On trajectory generation and active impedance for running trot-ting Dynamic Walking 2014
3. M. Focchi, V. Barasuol, I. Havoutis, J. Buchli, C. Semini and D. G. Caldwell,Local reflex generation for obstacle negotiation in quadrupedal locomotion,in Int. Conf. on Climbing and Walking Robots (CLAWAR), 2013.
4. A. Winkler, I. Havoutis, S. Bazeille, J. Ortiz, M. Focchi, D. Caldwell andC. Semini, Path planning with force-based foothold adaptation and virtualmodel control for torque controlled quadruped robots, in IEEE InternationalConference on Robotics and Automation (ICRA), 2014.
5. B. Rainer, J. Kurth, G. Schreiber, R. Koeppe, A. Albu-Schffer, A. Beyer,O. Eiberger, S. Haddadin, A. Stemmer, G. Grunwald and G. Hirzinger, TheKUKA-DLR lightweight robot arm - a new reference platform for roboticsresearch and manufacturing in robotics (isr), in 41st int. symposium on and2010 6th german conference on robotics (robotik), 2010.
6. http://www.hdtglobal.com/services/robotics/.7. Barrett technology: http://www.barrett.com/robot/products-arm.htm.8. W. T. Townsend and J. K. Salisbury, Robots and Biological Systems: Towards
a New Bionics? (1993).9. T. Boaventura, C. Semini, J. Buchli, M. Frigerio, M. Focchi and D. G. Cald-
well, Dynamic torque control of a hydraulic quadruped robot, in IEEE In-ternational Conference in Robotics and Automation, 2012.
10. S. Schaal, The SL simulation and real-time control software pack-age Technical Report, (Online) Accessed February 2011 at http://www-clmc.usc.edu/publications/S/schaal-TRSL.pdf, (2006).
11. M. Frigerio, J. Buchli and D. G. Caldwell, Code generation of algebraic quan-tities for robot controllers, in IEEE/RSJ International Conference on Intel-ligent Robots and Systems (IROS), 2012.
12. R. Featherstone, Rigid Body Dynamics Algorithms (Springer US, 2008).